Category Archives: Medication Informatics

UK Drug information, knowledge, coding and classification

I wrote this piece below to brief some US colleagues, but thought it might be worth sharing with a broader audience for what’ it’s worth.

The coding of drugs and medical devices in the UK is based on the NHS’s Dictionary of Medicines and Devices (dm+d) which is very closely related relevant sections of SNOMED (all dm+d identifiers are SNOMED codes).

It provides a hierarchical  representation of medicines starting from Virtual Therapeutic Moieties down to individual manufactures packs as illustrated in this diagram taken from the the dm+d website

dm+d also contains a “Trade Family” Class which is equivalent to the VTM class to allow the prescription of those small number of branded medicine where no VTM exists (eg Lipitor) which could otherwise only be prescribed at the  AMP or AMPP level which is at odds with usual inpatient prescribing practice which occurs at VTM level.  Trade Family TF and the associated Trade Family Group (TFG) and more fully described in the  SNOMED CT® UK Drug Extension Editorial Policy .

Despite it’s name medical devices coverage in dm+d is very limited with coverage by and large only extending to medical devices listed in the NHS Drug Tariff which includes all items including medicines, dressings, appliances, reagents and devices that can be prescribed on the NHS by UK general practitioners.

VTM, VMP,TF and TFG concepts exists as part of the core SNOMED namespace while VMPP and AMPP concepts form part of the UK extensions to SNOMED

The dm+d contains includes medicines by both generic name (usually aligned to International Nonproprietary Names ) and UK trade names and is designed to promote generic prescribing (the usual UK practice) for all medicines except in those few cases where generic prescribing is clinically contraindicated (even when only branded products are available to fill such prescriptions).

dm+d contains information about drug forms, strength, and licenced routes of administrations, but contains little other clinical information about the drug. However, it does provide mechanisms to link dm+d to a range of drug knowledge sources (e.g. the British National Formulary BNF) and many commercial drug knowledge suppliers maintain mappings from dm+d to their proprietary identifiers of have adopted dm+d as their native coding. These links allow prescribing based on dm+d to be link to a range of knowledge sources and prescribing decision support tools.

dm+d also contains a large number of data and metadata  items related to UK pricing, regulatory, best practice and supply chain operation (including links to the information in the Drug Tariff) which will be required by ePrescribing systems that wish to comply with UK statutory, regulatory and best practice recommendations and operate satisfactorily with medicines supply chain arrangements that apply in the NHS. There is a considerable volume of useful guidance on the implementation of dm+d freely available via the web with the dm+d web site providing a good start point from which to find this.

There are a numbers of sources of high quality UK specific drug information and knowledge to support both professional responsible for prescribing and administering medicines and patients which can be incorporated in to prescribing and medicines management systems. Information in these sources includes both eye-readable information (manufacturer’s data sheet (SPC’s), drug monographs, warnings, prescribing guidelines, patient information leaflets etc) available to the relevant actor at appropriate points in the treatment process and machine readable data and metadata intended to support automated prescribing, dispensing, administration and associated decision support. Such data will be provide in a structured and coded format to facilitate automated processing and may be supplied with computable  decision support rules and/or software components and APIs to aid it’s implementation.

Computable data is typically intended to support functions that might include.

  • Therapeutic recommendations
  • Drug-drug interactions
  • Drug-lab interactions
  • Drug contraindications
  • Dose recommendations/calculations
  • Dose range checking
  • Defined daily doses
  • Cross-sensitivity allergy checking
  • Drug administration
  • Side-effect identification
  • Drug substitutions
  • Drug-Gene interactions

Drug knowledge suppliers assemble their data from a range of primary and secondary sources. Each provides a different range of data which is likely to be provided in a proprietary format using a variety of open and proprietary coding systems. Common coding systems found in UK data sources include (Read, CTV3,  (widely used in UK general practice), dm+d, ICD, SNOMED (Op sit), ATC, BNF, PIP) mappings exist (of variable completeness and quality) between many of these systems where their domains overlap – Coding systems common in the US like NDC and LOINC are rarely found in UK drug knowledge sources (although some direct and indirect mappings may exist).

There are some significant differences in regulation and  prescribing practice,  reflected in the way medicines are supplied, between Europe and the USA, which affect  (while there are also differences between individual European countries there is increasing harmonisation across the EU and EEA) drug information requirements and ePrescribing and medicines management and administration systems.

  • Drug names both generic and trade names can be different for the same drug
  • The range of forms/strengths available can be different
  • Licensed indications can be different and some drugs available in the US  may not be available at all in the EU and vice versa
  • Legal classification are different and often not directly comparable
  • Most medicines in the EU are supplied and dispensed in patients packs (invariably blister packs for solid medications) – For drugs with a short duration of use these pack will be designed to provide one course of the medicine, for drugs with a longer or indeterminate period of use patients pack will generally be for 28 days supply, although other sizes of patient pack exist for some drugs.  Bulk packs of medicines designed to be dispensed into vials, compliance and unit dose systems and medicines pre-packed for unit-dose dispensing  are NOT  generally available.  This reflects the very low levels of unit-dose dispensing in UK hospitals compared to the US which is probably the single most significant difference in medicines management practice between US and UK hospitals.

Further Information

References to further information about issues covered are generally included as hyperlinks in the text above. Additional sources of note:

British National Formulary BNF – The UK “Gold Standard” for medicines information and guidance on prescribing.

Royal College of Physicians – Standards for medication and medical device records – technical annex  Contains and points to some of the latest work

NHS – dm+d Implementation Guide (Secondary Care) 2009, but still full of relevant, useful and detailed information


Medication Repository Anyone?

I was writing this piece when I read Ian McNicoll’s stimulating blog piece “EastendEHRs? – Dr Leggs’ Diary”   Ian talks about community medication records, very much the theme of this piece.

It seems to me that a shared medication record is the single most useful thing that could be provided in any health community and that while such a service is not without it challenges it is eminently doable.

I also feel that working through what such a service would look like and how it might be delivered could help us clarify our understanding about a number of technical, clinical and governance issues that would have general applicability with regard to the sharing of other parts of records or record summaries.

In my mind a medication repository is a logical construct, that might be physically instantiated in many ways to provide a single authoritative source with regard to medication information for a particular patient I envisage that all applications that need to read or write medication data would operate against the repository, or a locally maintained cache of it, rather than managing their own medication record.

Ideally, the repository should contain details of all medication from all care setting including patients self-medication, but a useful start could be made with a more limited scope (say just GP, outpatient, community and hospital discharge medication.)

There are many potential architectures for a medication repository. I am not necessarily suggesting a single medication repository for a health community (although this might be the most practical solution), there could be many (in the extreme, one for each individual) and they don’t have to have exclusive geographical scope (patients would be free to choose from all available service providers) the important things are that there is a single authoritative source under shared governance, a means of discovering where this is for a particular patient and that all repositories support a common API so that any application can easily work with multiple repositories.

Building appropriate governance arrangements around medication repositories is perhaps the most significant challenge, but also a particularly interesting one that will allow us to explore approaches in a constrained (and therefore manageable) scope that we can later apply to other aspects of the record.

Governance arrangements need to ensure that all those with rights to the information in the repository can be confident that their rights will be respected and protected for all time even in the event of the failure of a particular repository provider – I envisage some form of information escrow to achieve both of these things.

Governance arrangements also need to ensure that the provenance of information is secured, handle the allocation and transfer of clinical responsibility associated with information held in the repository. and the ongoing responsibility for management of a patients medication (changing cancelling adding medication.)

These governance issues illustrate the inseparability of the information and clinical governance issues, an inseparability which I believe applies much more broadly and is something not properly recognised in the implementation of existing systems.

Moving this idea forward requires some proof of concept work, which in turn requires someone willing to build an operate and experimental repository and a couple of application providers willing to work with it – ANY VOLUNTEERS.

EPS and the Electronic Transfer of Prescriptions

I first worked on the forerunner of what we now call EPS (the Electronic Prescription Service) in 1986, when I wrote an evaluation of a project to using a memory card to transfer prescriptions from GP practices to pharmacy for the then DHSS. However, even this was not quite the beginning of the EPS story as there had been some other projects earlier in the 1980s.

By 1986 most community pharmacies were already computerised (driven by a requirement for typewritten medicine labels and the widespread use of online order entry by the pharmaceutical wholesalers), GP computing was beginning to take off (although still at only about 5% penetration) the Prescription Pricing Authority was computerised and it already seemed obvious that it would be a good idea if prescriptions could flow electronically.

By the early 1990s GP computing had boomed and we had reached the point where 80% of prescriptions reaching community pharmacy were computer generated. With 600 million+ items prescribed annually it seemed an obvious “no-brainer” to try and move from a situation where much of the same information was typed and re-typed in to three different computer systems. The problem also didn’t look too difficult. Prescriptions consist of a small amount of fairly structured data that it would be simple to convert into a computable form that would not make excessive demands on the bandwidth or storage capacity of the then available technologies.

Numerous pilots ensued. These fell into two groups. Firstly those using some form of machine readable token, still physically transported from GP to pharmacy (this included chip cards, smart cards, magnetic stripes, 2D bar codes and optical cards). Secondly, there were those using online communication (either point-point between GP and pharmacy or via some form of email or relay service) There were a couple of attempts to launch a commercial service and I was involved in a project that tried to set up a joint industry owned service (it would have been like BACS but for prescriptions). These projects proved the concept and addressed many of the practical and professional issue involved and with the birth of the NPfIT it picked up the EPS baton in England to take forward what had been learned in this early work.

Image of Care Card
Care Card from 1987 Exmouth Smart Card Project

Twenty Five years after the first work, we have fairly comprehensive infrastructure in place in GP practices and community pharmacies in England and most have software to support EPS release 1, with about a third of prescriptions produced by GPs flowing to the EPS service. However, NHS Prescriptions (the successor organisation to the PPA) is still processing paper prescriptions (although in now uses scanning and OCR rather than manual data entry) and less than 2% of prescriptions are processed using EPS in pharmacies.

Why have we made so little progress in a quarter of a century?

There are a number of reasons, but paramount has been the determination by Government (driven by the pharmacy lobby) that EPS should not be allowed to become a disruptive technology, that could radically alter the medicines supply chain and potential destabilise community pharmacy. This in my view has been a critical error which has denied potentially significant benefits to patients, the NHS and those willing to take advantage of the commercial opportunities to drive efficiency and quality in the supply chain.

However, in the current economic climate

Government can no longer afford to take a protectionist stance and EPS release 2 provides the opportunity to improve both quality and efficiency, mainly by improving the truly dreadful process patients have to follow to get regular medication.

Patient concordance with medication is really poor (see my earlier blog post ) and in the medium term this increases costs, morbidity and reduces patient’s quality of life. EPS can make repeat prescribing more efficient and convenient but more importantly makes repeat dispensing (where the pharmacy takes over responsibility for refills) a much more practical option than in its current paper based form (indeed with EPS 2 repeat dispensing should completely replace repeat prescribing) While supply issues are far from the whole story with regard to concordance they are a big factor and a switch to pharmacy managed repeat dispensing enabled by EPS should have a significant positive impact, not to mention improved patient convenience and efficiency gains, particularly for GP practices.

EPS 2 has only just started to roll-out and achieving the benefits it promises requires action in a number of areas.

  • An acceleration of the role-out so that the benefits EPS2 offers “as is” can be realised. We should stop worrying about the commercial impact on community pharmacy and abandon the requirement for a Secretary of State’s Direction for local implementation.
  • We need to address those issues that mean not all prescriptions can use EPS.
  • We need to address the legitimate concerns of pharmacy with regard to business continuity in an EPS environment.
  • We need to address aspects of EPS that have a negative impact on efficiency in community pharmacy and ensure that future releases deliver the key benefits that pharmacy wants (including full automation of the application of exemptions and remissions from prescription charges and automated line-item reconciliation between dispensing and payment).

It’s been a long time in coming, we are nearly there and it would be good to see EPS becoming ubiquitous before I retire.

Medication Concordance

I’ve been doing some work concerned with patient concordance with medication in UK General Practice and have just completed a quick trawl through the literature.

I knew that patient concordance was not good, but what I see is truly worrying with data showing that  only about 40% are still taking the prescribed medication after 3 month across a range of treatments which one would expect to be long-term if not life long where poor concordance has serious sequelia.

People drop out at all stages of the process and while not all of the dropouts are easily visible to healthcare professionals much of it would be apparent from patient’s failure to request/collect repeat scripts or in the pharmacy where medicines dispensed go uncollected.

IT systems in both GP and pharmacy can easily identify these non-concordant patients but these does not seem to happen as much as might be expected, maybe in part at least because of the work involved in addressing the problem and the patchy success of attempts to do so reported in the literature.

The long-term cost of poor concordance in financial terms and in terms of morbidity and mortality is considerable and focused efforts at improving concordance should have a strong economic case. However, we see a problem here that inhibits many potentially quality improvements in the NHS namely that demonstrable cost savings don’t occur either in the budgets or timeframes that matter to those designing and delivering services. With the prospect of it doubling the drugs bill savings year hence or in someone else’s budget are not very alluring to commissioners.

Will things be different in the new world of GP Consortia? Maybe GPs commissioners  will be better placed to look at whole system costs, but I doubt they will welcome short-term cost hikes for long-term savings. Solving this problem, which affects not just medicines but a whole raft of technological innovations and non-technological service redesign, requires a new approach difficult to imagine in a cash-limited age of austerity. In the case of medicine maybe the industry can help with pricing approaches that reduce short-term cost in return for a share in long-term savings, but implementing such new approaches in the context of UK pharmaceutical price regulation is probably beyond the scope and ability of GP Consortia.

Existing IT systems in GP and pharmacy can help us identify non-concordance and we have evidence to inform interventions more likely to be effective. The Electronic Prescribing Service missed the opportunity to provide dispensing feedback to GP, but could be easily updated to do so and the Internet and mobile –devices offer new low-cost approaches to gain concordance. However, unless we create an environment that creates incentives to act I doubt we will make much progress.

I’m happy to share some of the literature once I have it consolidated, but for now here is a single paper which reviews much of the work